Research Associate (12 month term)
Core
Developing physics-inspired approaches to neural network compression and knowledge distillation to understand model scaling under constrained resources.
Role type
Research Associate (Physics & Machine Learning)
Builds
Theoretical frameworks, numerical experiments, and research software for neural network compression.
Domain
Physics (statistical physics, renormalization group) + Machine Learning
Deliverable
research
Required skills
statistical physics, renormalization group theory, machine learning, deep neural networks, Python, PyTorch, computer vision, statistical analysis, GPU computing
Preferred skills
knowledge distillation, information theory, neural architecture search, edge computing, scientific publications
Technologies
PyTorch, Python, Linux, Git
Responsibilities
Formulate mathematical frameworks for applying renormalization-group concepts to compression; Develop and evaluate ML models for compression; Design and conduct numerical experiments on model scaling; Develop research software and experimental pipelines; Analyze experimental results and uncertainty; Collaborate on scientific publications and open-source tools.
Seniority
Mid-level Researcher (PhD required)